|

|

Posts

(9 min read) In the early 2000s, I remember browsing the internet as something very top-notch and advanced. The internet was transitioning from Web 1.0 to Web 2.0, and this period was crucial in defining the business model behind the revenue. Should users pay to access content, or should content be free and supported by advertising? My curiosity about the history of online advertising arose from a desire to understand the current revenue model of the internet and where it might be headed. This article explores the evolution of online advertising, its impact on the internet, and its future trajectory. **The Birth of Online Advertising** The early days of the World Wide Web introduced cookie technology, crucial for online advertising. Cookies were created in the mid 1990s to keep the shopping cart list of early e-commerce applications, ensuring that selections were retained until purchase. Cookie technology was first integrated into the Netscape browser and later into other browsers. As websites began incorporating third-party elements, third-party cookies came into existence. Initially, there were no restrictions imposed on these third-party cookies, which allowed early advertisers to gather and monitor users' online activities. Before the introduction of cookies, the internet was a private space, but their implementation made it subject to monitoring (1). |Notes| |------------| |In 1994, cookie technology became part of the beta version of the Netscape browser. It was designed to keep track of transactions, like maintaining items in a shopping cart. Initially, cookies were accepted without notifying users. They are crucial for smooth online transactions. Despite of this, the use of third-party cookies allows data collection and aggregation, creating detailed profiles of internet users based on personal information and browsing habits. Some websites require both first-party and third-party cookies to be enabled for access. Without third-party cookies, these sites would need to add more features themselves and could miss out on creative services such as social networks and performance metrics offered by external parties. Therefore, this reliance on cookies needs further examination and regulation (1a).| Early experts on internet privacy issues began expressing concerns about third-party cookies. These cookies had the potential to track users' activities, thus by accident contributing to the emergence of the internet advertising industry, which significantly influenced the direction of the internet. Internet advertising was perceived as more cost-effective than traditional methods because it eliminated printing costs and was viewed as an information superhighway with limitless potential (2). **The Dot-Com Bust and Recovery** A 2001 article in the New York Times mentioned that the dot-com era was thought to be over, viewing early internet entrepreneurs as failed rock stars who would be replaced by more conservative and smart entrepreneurs (3). Online advertising declined because many players did not survive this wave of changes. Nevertheless, it started to recover around 2004, with the emergence of Web 2.0, which facilitated the buying and selling of advertising on websites. As a result, advertisers in the USA spent 10 billion dollar on websites in that year, accounting for about 14% of all advertising spend (4). In the mid 2000s, there was much discussion about how to generate revenue from internet services in the Web 2.0 era. Emerging services like the early version of Facebook prompted debates on whether users should pay for online services. Some reasons for users wanting to pay included the desire for services to evolve and survive, the belief that it was the correct thing to do, and concerns that advertising revenue was not stable. Conversely, reasons for not paying included the tradition of free services, funding from the advertising industry, and reluctance to use credit cards online (5). Business recommendations at the time suggested that users willing to pay for content would not represent a sustainable revenue source unless the content was rich and special. Casual users or those seeking content infrequently could be charged with more pricing flexibility, such as a pay-per-view service (6). **The Rise of Programmatic Advertising** Research during the initial stage of Web 2.0 shaped the evolution of online advertising in the following years. Traditional offline media like radio, TV, printed press, and billboards initially saw online advertising as a complement to their strategies. Nonetheless, the impact was inevitable, as printed advertising markets blamed online advertising for decreased ad revenues. Online advertising, a transformative force in the industry, began with search and display advertising. This distinction highlights the different nature of advertising sold by search engines versus publishers. Instead of buying ad space from each publisher one by one, innovations like programmatic advertising in the late 2000s made it easier for advertisers to connect with publishers using trading platforms. Programmatic advertising, a type of online advertising, simplified the process by automatically selecting the audience and placing the best-fitting ads from online ad spaces chosen through bidding. Despite this innovation, many people in the industry were slow to adopt programmatic advertising. This was because it needed new skills to understand the results and see if a campaign was successful. Many advertisers also said it felt like a "black box" because they didn't fully understand how it worked. |Notes| |------------| |Traditional advertising involves considerable manual effort to secure ad space and negotiate placements, which can be daunting given the growing demand in digital marketing. Programmatic advertising automates this process, enabling advertisers to manage ad placements in real-time through instantaneous buying of ad inventory. This approach offers enhanced insights and gathers data from multiple publishers, providing advertisers with clear visibility into campaign performance. Automation includes tools like Data Management Platforms (DMPs), which create customer profiles based on tastes, habits, and preferences stored as data cookies. Supply Side Platforms (SSPs) manage available ad space on advertising platforms, adjusting based on user channels. Demand Side Platforms (DSPs) connect publishers and advertisers, evaluating precise ad pricing through auction bidding. While programmatic advertising makes operations smoother, marketers may struggle with understanding the technical details needed to manage campaigns effectively and evaluate their performance (6a).| While these challenges persist, the landscape of online advertising is also being influenced by regulations such as the European GDPR, which may decisively change the industry's outlook. The full implications of the GDPR are yet to be fully understood by the market. Concerns have arisen regarding the legality of tools used for data sharing, such as analytics platforms. Still, advancements towards compliance with GDPR regulations are being made across the industry, which includes updates to analytics technologies to ensure alignment with these new standards (7). |Notes| |------------| |In 2023, the European Union issued substantial fines under GDPR rules to major tech firms for privacy breaches. Meta was fined €1.2 billion for moving large amounts of EU user data to the US. Additionally, Meta was fined €390 million for pushing users to agree to their data being used for personalised ads. TikTok was fined €350 million for mishandling children's accounts, which defaulted to public settings and allowed comments without proper protections. Moreover, TikTok was fined €14.5 million for processing data of children under 13 without parental consent and failing to disclose how children's data would be used. These fines show the EU's commitment to enforcing strict data protection rules across the tech industry (7a).| **Ad-Free Services and User Preferences** As of 2024, there is skepticism about whether customers are willing to pay for ad-free services. Some changes in social media, such as recent moves towards ad-free feeds, represent a new trend that needs further study. Streaming services offering unique content have begun lowering subscription fees. Consumers willing to receive ads instead of paying a subscription often prefer ads tailored to their interests. This raises important questions about security and privacy, as tracking user preferences extends beyond traditional limits. Additionally, it prompts us to reconsider the effectiveness of defined target audiences for advertising and whether they successfully engage users (8). **The Challenge of Targeting Audiences** We live in a world where advertising is everywhere: social networks, mobile notifications, streaming services, mailboxes, etc. Advertising specialists create target audiences, but in the era of abundant online advertising, ad skipping behaviour may indicate that audience identification is becoming more challenging. An Austrian experiment asked an agency to design broad-, narrow-, and no-targeting strategies in a fictitious Facebook campaign. The results illustrated the importance of targeting feasibility. Targeting is better than no targeting, but excessively narrow targeting can lead to higher costs and lower reach, making it unprofitable. This suggests that fulfilling users' desires for custom and tailored advertising can be profitable but inevitably involves targeting some wrong people. The Austrian experiment also showed an inverse correlation between reach and click-through rate (CTR), indicating that maximising both objectives simultaneously is not possible. |Notes| |------------| |The recent introduction of Apple's tracking transparency framework, which requires user consent for third-party cookies, has decreased the quality of data for creating audiences on this platform. While this poses a challenge for targeting audiences, it is positive news for user privacy and security (8a).| The online advertising industry is complex and difficult to analyse from an academic perspective. That said, the professionalism within the sector provides opportunities to enhance the quality of advertising and its impact on audiences without compromising the industry. Understanding current challenges, such as the annoyance factor, is crucial. Aggressive online marketing techniques involving repetition and intrusiveness can negatively affect the advertised product and brand. Factors to consider, regardless of user age, include the lack of personalisation, ads occupying at least 50% of the screen, fixed-size or non-skippable ads, and pop-up ads. Native advertising, which is sometimes not recognised as ads, creates a less intrusive experience and leads to less annoyance (9). **The Role of Artificial Intelligence** Given the widespread use of artificial intelligence, an interesting question arises: Can AI solve the current online advertising challenges? The emerging AI industry, projected to be a 13 trillion dollar ecosystem by 2030, will inevitably transform all digital and non-digital industries. In online advertising, AI advancements are expected to enhance data analysis, copywriting, video production, and the creation of recyclable advertising concepts. Recyclable advertising can adapt to different contexts, platforms, or devices, changing aspects of publicity to better target audiences. In Internet 3.0, which includes the Internet of Things, devices that understand our needs could suggest highly focused advertising. Yet, this prompts questions about the boundaries of advertising and its potential to integrate into our lives without being intrusive or annoying, all while adhering to ethical and regulatory standards (10). **Conclusions** The early online advertising industry has quickly developed into a smarter industry with specialised and advanced tools from tech giants like Google and Meta, shaping the advertising landscape. Even so, issues in the industry are leading to regulatory consequences and changes in how applications will use online advertising in the future. More research is needed on the impacts of advertising on the public and the outcomes of these issues, particularly concerning ad skipping, the accuracy of advertising proposals, and the perceived value of free applications balanced with viewing advertisements. The effects of regulation and legal cases will soon become visible. **Sources** (1) Dean Donaldson (2008): Online advertising history Consulted on 24/06/2024 URL: [Link](https://nothingtohide.us/wp-content/uploads/2008/01/dd_unit-1_online_advertsing_history.pdf) (1a) Ian D. Mitchell (2012): Third-party tracking cookies and data privacy Consulted on 24/06/2024 URL: [Link](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2058326) (2) Karrie Chen et al. (1999): The history of advertising Consulted on 24/06/2024 URL: [Link](https://kccesl.tripod.com/ESL91NetProject.grprojs.html) (3) John Schwartz (2001): When the internet was, um, over? Consulted on 24/06/2024 URL: [Link](https://www.nytimes.com/2001/11/25/style/dot-com-is-dot-gone-and-the-dream-with-it.html) (4) Masoud Nosrati et al. (2013): Internet Marketing or Modern Advertising! How? Why? Consulted on 24/06/2024 URL: [Link](https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=40916156a90bbee1eea9cccb33d3c6fd3a7fa8f3) (5): L. Richard Ye et al. (2004): Fee-Based Online Services: Exploring consumers' willingness to pay Consulted on 24/06/2024 URL: [Link](https://scholarworks.lib.csusb.edu/jitim/vol13/iss1/12/) (6) Cheng Lu Want et al. (2005): Subscription to fee-based online services: What makes consumer pay for online content? Consulted on 24/06/2024 URL: [Link](http://ojs.jecr.org/jecr/sites/default/files/paper4_20.pdf) (6a) K Uday Kiran et al (2024): Role of programmatic advertising on effective digital promotion strategy: A conceptual framework Consulted on 24/06/2024 URL: [Link](https://iopscience.iop.org/article/10.1088/1742-6596/1716/1/012032/pdf) (7) Lucian Blaga (2022): Online advertising - History, evolution and challenges Consulted on 24/06/2024 URL: [Link](http://economice.ulbsibiu.ro/revista.economica/archive/74311ungureanu&popescu.pdf) (7a) Niall McCarthy (2024): The Biggest GDPR Fines of 2023 Consulted on 24/06/2024 URL: [Link](https://www.eqs.com/compliance-blog/biggest-gdpr-fines/) (8) Charles R. Taylor (2024): Ad skipping, the "ad free internet" and privacy: a call for research Consulted on 24/06/2024 URL: [Link](https://www.tandfonline.com/doi/full/10.1080/02650487.2024.2309747) (8a) Iman Ahmadi et al. 2024): Overwhelming targeting options: Selecting audience segments for online advertising Consulted on 24/06/2024 URL: [Link](https://www.sciencedirect.com/science/article/pii/S0167811623000502?via%3Dihub) (9) Maria Rigou et al. (2023): Online Ads Annoyance Factor: A Survey of Computer Science Students Consulted on 24/06/2024 URL: [Link](https://www.intechopen.com/online-first/86252) (10) Eyice Başev, S. (2024): The role of artificial intelligence (AI) in the future of advertising industry: Applications and examples of AI in advertising Consulted on 24/06/2024 URL: [Link](https://www.ijetsar.com/Makaleler/1386856240_6.%20167-183%20Sinem%20Eyice%20Ba%C5%9Fev.pdf)

|

(8 min read) The first time I learned about organising and processing business data was during an "Analysis and System Design" course. It introduced the concept of designing a system from business requirements to interface, focusing on Transactional Systems. Additionally, the course taught a methodology to create a data structure for further data analysis and visualisation, applicable to Decision Support Systems, which formed the foundation of what was referred to as conventional data analytics and business intelligence before the rise of artificial intelligence in data analytics. **The Challenges of Data Analytics** In the early 2000s, conventional data analytics was groundbreaking and disruptive, but academic publications rarely mentioned terms like Business Intelligence, Business Analytics, or Big Data. By 2010, these terms had become exponentially popular (1). That year, 2.4 trillion dollar was spent on software services in the business sector. Despite this investment, only 32% of technology projects were successful, meaning 68% failed to deliver value to the organisation (2). Successful projects typically achieve this by ensuring alignment between the solutions and the business strategy. Conversely, misalignment between these can also lead to project failures. Data analytics projects aim to deliver business value, prioritising practical solutions over perfect ones. Data scientists typically spend between 75% and 80% of their time cleaning, organising, and preparing data for analysis. Therefore, ensuring that the data is well-prepared and meets the business's needs greatly improves the chances of achieving successful outcomes from analytics projects. The results of this effort are often presented in dashboards with KPIs for performance, frequently using descriptive and diagnostic analytics (3). This process is time-consuming and challenging, especially when bridging the gap between the objectives of business users and the requirements for data visualisation. Every company has different levels of data processing and development capabilities, and aligning these capabilities with business requirements is crucial. When there is a mismatch, there is an opportunity to enhance the company's capabilities if aligned with the business strategy. Nonetheless, data analytics is generally a daunting task, and results can become obsolete rapidly. Evolving organisational needs, strategic changes due to competition, mergers, acquisitions, and the emergence of new market players can demand new data analytics capabilities for rapid decision-making. This impacts the feasibility and longevity of data analytics projects, as dynamic organisations find it difficult to manage the cost of changes, undermining their confidence in data analytics as a reliable business solution. **The Advent of Artificial Intelligence** Artificial intelligence is not a new concept. The first neural network was created in 1950 by Marvin Minsky and Dean Edmonds at Harvard University, simulating 40 neurons on a computer (4). Nowadays, over 80% of organisations see AI as a strategic opportunity, with nearly 85% viewing it as a way to gain a competitive advantage (5). In the early 2010s, conventional data analytics dominated the field. Nevertheless, AI's emergence was expected to make data analytics more powerful, addressing difficulties unexplained by conventional descriptive and diagnostic analytics. AI and machine learning enhanced new aspects of data analytics, such as predictive and prescriptive analytics, automating data preparation, unification, and organisation. Additionally, AI could automate some parts of the code needed for conventional data analytics projects, reducing errors and accelerating development, thus helping organisations remain competitive. By analysing historical data and current business priorities, AI can suggest actions to enhance company performance (6). |Notes| |------------| |Artificial Intelligence (AI) is transforming various sectors. For example, in education, AI analyses student data, including scores and other relevant information, to suggest improvements in learning methods and automate tasks such as tracking student performance. In retail, AI improves demand forecasting and personalises product recommendations to enhance inventory management. In manufacturing, AI predicts when machines need maintenance and oversees factory operations through image recognition. In healthcare, AI assists in diagnosing conditions and allocates resources for patient care. In telecommunications, AI optimises service quality through maintenance and operational efficiencies. In banking, AI enhances customer service by efficiently routing calls, detects fraud, and customises financial assessments for credit eligibility.| **Barriers to AI Adoption** Despite its potential benefits, AI adoption faces several challenges. As of 2022, only 8% of companies in the EU used at least one AI technology. Early adopters of AI have applied it to stock market analysis, valuing real options, and fraud detection. Smaller companies are less likely to adopt AI. For instance, in Austria, 92% of small companies, 85% of medium-sized companies, and 74% of large companies have not yet considered using AI. Common challenges include high development costs, lack of skilled staff, management and legal risks, and ineffective data management practices (7). These data management challenges, a barrier to AI adoption, are connected with issues in implementing conventional data analytics projects. If an organisation cannot effectively implement conventional data analytics projects and foster a data culture, it is unlikely to succeed in implementing advanced data analytics projects, where AI plays a crucial role. A strong indicator of a mature data culture is the industrialisation of conventional data analytics projects, where the company routinely reuses procedures and development components as similar cases arise. This standardises knowledge and increases production, indicating that the company has achieved a level of stability necessary for implementing advanced data analytics projects powered by AI (8). **Generative AI as part of the solution** Organisations that use third-party IT infrastructure often consume cloud solutions from providers like Amazon, Google, and Microsoft. These providers offer three categories of services: Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS). Not surprisingly, they are integrating generative AI into their services. For example, Microsoft's 11 billion dollar investment since 2019 in an exclusive partnership with OpenAI illustrates how cloud providers are enhancing their services with generative AI. |Notes| |------------| |Generative AI uses advanced techniques in natural language processing (NLP) and machine learning to create different types of content, such as text, images, audio, and synthetic data, based on prompts or questions. It can be used for many purposes, like writing articles, powering chatbots, creating art and designs, composing music, generating speech, and making synthetic data for training and testing AI models.| Combining these services allows cloud-based enterprises to leverage IT solutions more effectively. Generative AI algorithms can interact with the cloud in a more human-like manner, making data interaction, access, and outputs more intuitive. Additionally, using the cloud infrastructure to create AI models offers benefits such as scaling computing resources up or down depending on the needs, providing flexibility and cost efficiencies. This adaptability is particularly advantageous for clients keen on using AI to improve the industries they operate in. By integrating generative AI with cloud services, organisations can optimise their operations and maintain competitiveness in a rapidly evolving technological landscape (9). **Data Preparation for enhanced Business Intelligence powered by AI** A typical AI business case often includes a clear application description, how the company will use it, the required data, and an AI expert to identify the best-fit algorithm. Pursuing an AI case incompatible with AI techniques can lead to failure, with non-existent, uncertain, or unpredictable results. Can a company use AI technologies if it has varying data quality? Even if the data is not clean or is difficult to process, which is common in companies with emerging data cultures, AI technologies can still be valuable. Both Generative AI and machine learning models can be combined to assist in cleaning processes, standardising, and labelling unstructured data. While assessing quality after these steps can be complex, the use of AI can significantly enhance the overall data analytics efforts and improve project outcomes. This effort can increase the initial cost of AI-driven projects, especially in cleaning and organising data, which is incomplete or inconsistent (10). Compared to the conventional data analytics era, this is advantageous because AI can handle and process more varied data types. Although more unstructured and incoherent data may require additional resources and costs, and extend the time before AI projects start delivering value, the long-term benefits of leveraging AI technologies make it a worthwhile investment. |Notes| |------------| |A business case highlighting the importance of data culture in one of Germany's largest utility firms demonstrates that a key factor behind the firm's AI maturity is its data-driven approach. This includes efforts to make the workforce see data as a valuable resource and the development of a digital infrastructure that supports strategic changes. These changes enable employees to participate in the data revolution as ambassadors of AI use cases that add value to the organisation. The spread of data culture within the firm has led to successful AI project implementations, supported by 3,000 rules that define good quality data, making it more visible and objective. The automation of digital and data processes is evident in three key data-driven decision support systems: real-time operations for performance stability, maintenance, replacement and repair of components, and long-term asset management. As more decisions become automated, clear responsibilities for decisions ensure accountability and organisation (10a).| **Conclusion** The transition from traditional data analytics to advanced data analytics powered by artificial intelligence presents significant opportunities for organisations. AI can automate many operational tasks, offering improved value and efficiency. Additionally, the emergence of generative AI models and their integration with cloud services can enable unprecedented scenarios for delivering value from data analytics projects faster and more powerfully than ever before. However, successful AI adoption still requires a robust data culture within companies. While advanced technologies can mitigate challenges associated with an underdeveloped data culture, establishing a strong data culture is crucial, even before developing AI solutions. By fostering this culture, businesses can fully leverage AI to achieve their strategic goals and remain competitive in a rapidly evolving technological environment. **Sources** (1) Tuncay Bayrak (2015), A review of Business Analytics: A Business Enabler or Another Passing Fad Consulted on 22/06/2024 URL: [Link](https://www.sciencedirect.com/science/article/pii/S1877042815038331) (2) Dennis et al. (2012), System Analysis and Design (3) Ralph Schroeder (2015), Big data business models: Challenges and opportunities Consulted on 22/06/2024 URL: [Link](https://www.tandfonline.com/doi/full/10.1080/23311886.2016.1166924) (4) Chandeepa Dissanayake (2021), Artificial Intelligence, a brief overview of the discipline Consulted on 22/06/2024 URL: [Link](https://www.researchgate.net/profile/Chandeepa-Dissanayake-2/publication/368852628_Artificial_Intelligence_-_A_Brief_Overview_of_the_Discipline/links/63fe13550d98a97717c5ba9d/Artificial-Intelligence-A-Brief-Overview-of-the-Discipline.pdf) (5) Ida Merete Enholm et al. (2021), Artificial Intelligence and Business Value: a Literature Review Consulted on 22/06/2024 URL: [Link](https://link.springer.com/article/10.1007/s10796-021-10186-w?trk=public_post_comment-text) (6) Mariya Yao et al. (2018), Applied Artificial Intelligence, a handbook for business leaders (6a) David Oyekunle et al. (2024): Digital Transformation Potential: The role of Artificial Intelligence in Business Consulted on 22/06/2024 URL: [Link](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4806733) (7) Rudolf Grünbichler et al. (2023), Implementation barriers of artificial intelligence in companies Consulted on 22/06/2024 URL: [Link](https://www.researchgate.net/profile/Gruenbichler-Rudolf/publication/371958928_IMPLEMENTATION_BARRIERS_OF_ARTIFICIAL_INTELLIGENCE_IN_COMPANIES/links/649ed4abb9ed6874a5eb4517/IMPLEMENTATION-BARRIERS-OF-ARTIFICIAL-INTELLIGENCE-IN-COMPANIES.pdf) (8) Mathieu Bérubé et al. (2021), Barriers to the Implementation of AI in Organisations: Findings from a Delphi Study Consulted on 23/06/2024 URL: [Link](https://scholarspace.manoa.hawaii.edu/items/1305e043-f68e-4485-bf7a-49e1e55c33ee) (9) Christophe Carugati et al. (2023), The competitive relationship between cloud computing and generative AI Consulted on 23/06/2024 URL: [Link](https://www.jstor.org/stable/resrep55201?seq=4) (10) Sulaiman Abdallah Alsheibani et al. (2020), Winning AI Strategy: Six-Steps to create value from Artificial Intelligence Consulted on 23/06/2024 URL: [Link](https://core.ac.uk/download/pdf/326836031.pdf) (10a) Philipp Staudt et al. (2024), How a Utility Company Established a Corporate Data Culture for Data-Driven Decision Making Consulted on 23/06/2024 URL: [Link](https://aisel.aisnet.org/cgi/viewcontent.cgi?article=1582&context=misqe)

|